Image Super-Resolution with Fast Approximate Convolutional Sparse Coding

Christian Osendorfer, Hubert Soyer, Patrick van der Smagt

Abstract

We present a computationally efficient architecture for image super-resolution that achieves state-of-the-art results on images with large spatial extend. Apart from utilizing Convolutional Neural Networks, our approach leverages recent advances in fast approximate inference for sparse coding. We empirically show that upsampling methods work much better on latent representations than in the original spatial domain. Our experiments indicate that the proposed architecture can serve as a basis for additional future improvements in image super-resolution.
BibTeX
@inproceedings{osendorfer2014image,
  author = {Christian Osendorfer and Hubert Soyer and Patrick van der Smagt},
  title = {{Image Super-Resolution with Fast Approximate Convolutional Sparse Coding}},
  booktitle = {ICONIP},
  year = {2014},
  doi = {10.1007/978-3-319-12643-2\_31},
}

Loading related content...

Figure